RR 145 - Reaction Inhibition in the Control of Exothermic Runaway · Title RR 145 - Reaction...

APPENDIX A

PHI TEC ADIABATIC CALORIMETRY DATA

PHI TEC ADIABATIC CALORIMETER

The PhiTec II is a low thermal mass adiabatic 100 ml calorimeter that can be used to

study the potential hazard of reaction systems by operating with no nominal heat

transfer to the surroundings. There is a built in pressure compensation feedback device

to maintain the integrity of the calorimeter can. Temperature, pressures and time are

logged throughout the test. Data can be analysed to obtain kinetic data. Adiabatic

calorimeters are often used to determine the potential hazard of chemical reactions.

The calorimeter was charged with 60 g of styrene and 0.44 g of benzoyl peroxide

initiator. The calorimeter was used in heat-wait-search mode which heats the can and

contents in stages until an exothermic reaction is detected. This occurred for the criteria

heat generation rate selected at 77°C. Thereafter the temperature and pressure generated

are tracked by the instrument with zero nominal heat transfer to the surroundings.

Figure A1 Runaway Polymerisation of Styrene with Benzoyl Peroxide Initiator

(Phi Tec HWS test)

-1

0

1

2

3

4

5

6

ln d

T/d

t

Rate of Temperature Rise (Pc46)

-2.9 -2.7 -2.5 -2.3 -2.1 -1.9 -1.7 -1.5

-1000/T

Figure A2 Activation energy & pre-exponential factor from Phi Tec PC46

-6

-5

-4

-3

-2

-1

0

1

2

3

4

ln((

dX

c /d

t)/(

1-X

c ))

a/R

PC46

Intercept = ln A

Gradient = E

-2.9 -2.8 -2.7 -2.6 -2.5 -2.4 -2.3 -2.2 -2.1 -2

-1000/T

APPENDIX B LABORATORY REACTOR STUDIES INHIBITION OF STYRENE POLYMERISATION

EXPERIMENTAL PROCEDURE

The aim of the experiments is to determine whether tert. butylcatechol can successfully inhibit

the runaway polymerisation of styrene under various conditions. The effect of the degree of

mixing on inhibitor effectiveness is also studied in brief. The inhibitor concentration is

expressed as a molar percentage and the initiator concentration as a percentage of the total mass

of reactant.

The required amount of tert. butylcatechol inhibitor is mixed with a small amount of styrene

and drawn into the double-acting air cylinder. The reactor is isolated and benzoyl peroxide

initiator loaded directly into the glass sight vessel. The remaining styrene is then loaded into the

feed vessel and heated to the required initiation temperature. On reaching this temperature,

styrene is passed into the glass sight vessel where it mixes with the initiator. When the initiator

has dissolved in the styrene, both reactants flow into the reactor. The reactor jacket is

maintained at a constant inlet temperature during the experiment. When the reactor set

temperature is reached, the air cylinder operates automatically and inhibitor solution is passed

into the reactor. A check valve prevents return flow from the reactor.

Experiments are performed either with an open, or an initially closed, system. In the latter case,

the reactor vent valve is closed and switched to automatic control after charging of the reactor is

complete. The pressure vent controller is set to open the reactor vent valve at the desired set

pressure. Temperatures and pressures are recorded by both fast and slow data logging systems

during the reaction. The reaction is also observed remotely and recorded by video system.

The conditions of the experiments and a summary of the main results are given in Table B1.

Graphs showing transducer reading versus time histories for key transducers in the reactor are

given for each experiment as Figures B1 to B16.

Table B1 Laboratory reactor experimental conditions and main results

Experiment number Lb26 Lb27 Lb28 Lb29 Lb30 Lb31 Lb32 Lb33

Test Conditions

Reactor volume (litres) 1 1 1 1 1 1 1 1

Benzoyl peroxide (%wt/wt)) 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

Tert. Butyl catechol (%m/m) 1 1 1 1 1 1 1 1

Reactor jacket temperature (°C) 100 110 120 120 120 120 120 120

Agitator speed (rpm) 300 300 300 300 300 300 Off 30

Inhibitor injection temperature (°C) 110 120 130 130 130 130 130 130

Relief valve set pressure (bara) 2 3 3 Open 3 Open 2 2

Max pre - injection temp. rate 3.7 10.3 15.5 11.6 21.9 23.3 30.0 27.1

(°C/min)

Temp. rate at injection (°C/min) 2.8 6.7 11.7 10.9 12.4 12.8 16.8 15.3

Max temp. rate after injection -ve -ve -ve -ve -ve 12.9 29.3 33.9

(°C/min)

Max Temp. (°C) 109.9 119.5 134.5 131.8 133.2 137.9 168.4 172.2

Overtemp. (°C) - - 4.5 1.8 3.2 7.9 38.4 42.2

Pre - injection pressure (bara) 1.22 1.26 1.04 0.97 1.35 0.96 1.18 1.38

Max pre - injection press rate 0.05 0.13 0.89 0.47 0.18 1.18 0.26 0.57

(bar/min)

Press rate at injection (bar/min) 0.32 0.66 0.44 1.97 0.26 1.08 0.11 0.64

Max. press rate after injection 2.62 1.31 1.90 0.84 0.35 1.45 2.92 2.16

(bar/min)

Max. press. (bara) 1.29 1.35 1.15 0.98 1.39 0.99 2.04 2.03

Times from initiation

Time to inhibitor solution Injection (s) 606.4 256.8 196.6 261.3 258.4 486.9 260.3 256.8

Time to max temp. (s) 607.7 259.3 210.6 269.7 267.8 518.8 299.9 284.3

Time to max press. (s) 640.8 413.2 202.6 490.7 269.5 520.0 298.5 284.6

Experiment 26

Critical variables

Reactor jacket temp: 100 °C

Agitator speed: 300 RPM

Inhibitor injection temp: 110 °C

Relief valve set pressure: 2 Bara

Run No:lb26 Date: 01 November 2000 File:lb26s.opj

70

75

80

85

90

95

100

105

110

115

Jacket In (RTD1)

Jacket Out (RTD2)

Vapour (tc1)

Liquid (tc2)

Tem

pera

ture

(°C

)

Pre

ssure

(bara

) Tem

pera

ture

(°C

)

0 500 1000 1500 2000 2500 3000 3500

Time (s)

Figure B1 Overall reactor temperature records

Run No:lb26 Date: 01 November 2000 File:lb26f.opj

Inhibitor solution

injected

Vapour (tc1)

Liquid (tc4) Liquid (tc2)

Pressure (P1)

1.4

110

1.3 105

100 1.2

95

1.1

90

1.0 85

80 0.9400 450 500 550 600 650 700 750 800

Time (s)

Figure B2 Temperature and pressure profiles in the reactor during inhibition

Experiment 27

Critical variables






85

90

95

100

105

110

115

120

125

Jacket In (RTD1) Jacket Out (RTD2)

Vapour (tc1)

Liquid (tc2)

Tem

pera

ture

(°C

)

0 1000 2000 3000 4000

Time (s)



85

90

95

Liquid (tc5)

Liquid (tc2)

injected

100

105

110

115

120

Tem

pera

ture

(°C

)

0.9

1.0

1.1

1.2

1.3

1.4

Vapour (tc1)

Pressure (P1)

Inhibitor solution

Pre

ssure

(bara

)

50 100 150 200 250 300 350 400 450

Time (s)


Experiment 28

Critical variables






90

100

110

120

130

140

Vapour (tc1)

Liquid (tc2 & tc4)

Tem

pera

ture

(°C

) P

ressure

(bara

)

0 500 1000 1500 2000 2500 3000 3500

Time (s)



140

injected

Liquid (tc2)

Liquid (tc4)

Vapour (tc1)

Pressure (P1)

Tem

pera

ture

(°C

)

Inhibitor solution

1.4

130 1.3

1201.2

1101.1

100

1.0

90

0.90 50 100 150 200 250 300 350 400

Time (s)


Experiment 29

Critical variables




Relief valve set pressure: Open test


90

100

110

120

130

Jacket Out (RTD2) Jacket In (RTD1)

Liquid (tc2)

Vapour (tc1)

Tem

pera

ture

(°C

) P

ressure

(bara

)

0 2000 4000 6000 8000 10000

Time (s)



Pressure(P1)

Vapour (tc1)

Liquid (tc2)

Tem

pera

ture

(°C

)

Inhibitor solution

injected

1.4

130

1.3

120

1.2

1101.1

1001.0

90 0.950 100 150 200 250 300 350 400 450

Time (s)


Experiment 30

Critical variables






80

90

100

110

120

130

140

Jacket Out (RTD2)

Jacket In (RTD1)

Vapour (tc1)

Liquid (tc2)

Tem

pera

ture

(°C

)

0 500 1000 1500 2000 2500 3000

Time (s)


Pre

ssure

(bara

) Tem

pera

ture

(°C

)

Run No:lb30 Date: 20 November 2000 File:lb30f.opj 140

Vapour (tc1)

Pressure(P1)

Liquid (tc2)

Inhibitor solution

injected

1.4

1301.3

120

1.2

110

1.1 100

901.0

80

0.950 100 150 200 250 300 350 400 450

Time (s)


Experiment 31

Critical variables




Relief valve set pressure: Open test

Run No:lb31 Date: 06 December 2000 File:lb31s.opj

90

100

110

120

130

140

Jacket In (RTD1) Jacket Out (RTD2)

Vapour (tc1)

Liquid (tc2)

Tem

pera

ture

(°C

) P

ressure

(bara

) Tem

pera

ture

(°C

)

0 1000 2000 3000 4000 5000 6000

Time (s)


Run No:lb31 Date: 06 December 2000 File:lb31f.opj

Vapour (tc1)

Pressure(P1)

Inhibitor solution

injected

Liquid (tc2)

1.4

140

1.3 130

120 1.2

110

1.1

100

1.0 90

80 0.9300 350 400 450 500 550 600 650 700

Time (s)


Experiment 32

Critical variables






20

40

60

80

100

120

140

160

Catch tank low (tc6)

Catch tank high (tc3)

Liquid (tc2)

Vapour (tc1)

Jacket Out (RTD2)

Jacket In (RTD1)

Tem

pera

ture

(°C

) Tem

pera

ture

(°C

)

0 200 400 600 800 1000

Time (s)

Figure B13 Overall reactor and catch tank temperature records


180

Relief vent valve opened

Pressure(P1)

Inhibitor solution

injected

Vapour (tc1)

Liquid (tc2)

2.0

1601.8

Pressure140 1.6

1.4 120

1.2

100

1.0

8050 100 150 200 250 300 350 400 450

Time (s)


Experiment 33

Critical variables






20

40

60

80

100

120

140

160

Catch tank low (tc6)

Catch tank high (tc3)

Liquid (tc2)

Vapour (tc1)

Jacket Out (RTD2) Jacket In (RTD1)

Tem

pera

ture

(°C

)

0 200 400 600 800 1000 1200 1400 1600

Time (s)

Figure B15 Overall reactor and catch tank temperature records


80

Inhibitor solution

100

120

140

160

Tem

pera

ture

(°C

)

1.0

1.2

1.4

1.6

1.8

2.0

Relief vent

valve opened

Pressure(P1)

injected

Vapour (tc1)

Liquid (tc2)

Pre

ssure

(bara

)

50 100 150 200 250 300 350 400 450

Time (s)


APPENDIX C PILOT REACTOR STUDIES INHIBITION OF STYRENE POLYMERISATION

EXPERIMENTAL PROCEDURE

The benzoyl peroxide initiator was pre-weighed in the laboratory, then loaded into the glass

feed column. The pre-prepared inhibitor solution was then charged to the injector vessel from a

drum installed on a weigh scale. The injector vessel was then pressurised to 15 barg with

nitrogen. The appropriate quantities of styrene were charged to the reactor and feed vessel 1

respectively. When the required initial temperature was reached in both the feed and reactor

vessels, styrene was passed from feed vessel 1 into the feed column allowing the benzoyl

peroxide initiator to be flushed through into the reactor. The reactor was then sealed. Heat

transfer fluid was passed through the jacket at a constant flow rate and the temperature and

pressure in the reactor were monitored during the course of the runaway reaction. Video

recording was used to observe the various reactants passing into the reactor. In experiments 1

and 2, agitation continued throughout the tests. For the final experiment 3, agitation of the

reactor contents was stopped automatically when a pre-selected temperature was reached. This

allowed settling of the reactor contents. In all experiments, the valve between the injector

vessel and the reactor was automatically opened on reaching a second pre-selected temperature,

allowing injection of the inhibitor solution. The reactor temperature and pressure were

continually monitored to ensure that inhibition of the reaction had been successful and that

runaway polymerisation did not resume. Reactor contents were then allowed to cool while

maintaining a constant jacket temperature. At the end of each experiment, the contents of the

reactor were transferred to a storage vessel and weighed.

Test conditions were chosen which provided respectable self-heating rates at the start of

runaway, while still allowing time for the successful inhibition of the reaction.

The flow rate of water to the reactor jacket was maintained at 20kg min-1 for each experiment.

The conditions of the experiments and a summary of the main results are given in Table 5.2 of

the main report. Graphs showing transducer reading versus time histories for key transducers in

the reactor are given for each experiment as Figures C1 to C12.

Experiment 1

Critical variables



Run No: P55 Date: 05 February 2001 File: P55s.opj

70

80

90

100

110

120

130

Jacket In (RTD2A)

Jacket Out (RTD9A)

Vapour (tc1)

Liquid (tc6)

Tem

pera

ture

(°C

)

0 1000 2000 3000 4000 5000 6000

Time (s)

Figure C1 Overall reactor temperature records


70

80

90

100

110

120

130

Viscosity

Liquid temp. (tc6)

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0 V

iscosity (cP

)

1000 2000 3000 4000 5000 6000

Time (s)

Figure C2 Reactor liquid temperature and viscosity profiles

0

Run No: P55 Date: 05 February 2001 File: P55f.opj

70

80

90

100

110

120

130

Vapour temp. (tc1)

Pressure (P2)

Liquid temp. (tc6)

Inhibitor injected

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pre

ssure

(bara

)

600 700 800 900 1000 1100 1200 1300

Time (s)

Figure C3 Temperature and pressure profiles in the reactor during inhibition

Run No: P57 Date: 03 May 2001 File: P57f.opj

70

80

90

tc6 tc4 tc3

tc2

tc1

0

5

10

15

100

110

120

130

Vapour temps.

Liquid temps.

Inhibitor injector

pressure (P5)

Tem

pera

ture

(°C

)

Pre

ssure

(barg

)

850 900 950

Time (s)

Figure C4 Reactor temperatures and inhibitor injection vessel pressure profile

Experiment 2

Critical variables

Agitator speed: 51.5 RPM



70

80

90

100

110

120

130

Vapour (tc1)

Jacket In (RTD2A)

Jacket Out (RTD9A)

Liquid (tc6)

Tem

pera

ture

(°C

)

0 1000 2000 3000 4000

Time (s)



70

80

90

100

110

120

130

Liquid temp. (tc6)

Viscosity

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0

Vis

cosity (cP

)

1000 2000 3000 4000 5000

Time (s)


5000

0

Run No: P56 Date: 12 February 2001 File: P56f.opj

70

80

90

100

110

120

130

Inhibitor injected

Vapour temp. (tc1) Pressure (P2)

Liquid temp. (tc6)

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pre

ssure

(bara

)

800 900 1000 1100 1200

Time (s)



70

80

90

0

5

10

15

100

110

120

130

tc6 tc4 tc3

tc2

tc1

Vapour temps.

Liquid temps.

Inhibitor injector

pressure (P5)

Tem

pera

ture

(°C

)

Pre

ssure

(barg

)

900 950 1000 1050

Time (s)


Experiment 3

Critical variables



Agitation stopped @: 119.5 °C

Run No: P57 Date: 03 May 2001 File: P57s.opj

70

80

90

100

110

120

130

Vapour (tc1)

Jacket In (RTD2A)

Jacket Out (RTD9A)

Liquid (tc6)

Tem

pera

ture

(°C

)

0 1000 2000 3000

Time (s)


Run No: P57 Date: 03 May 2001 File: P57s.opj

70

80

90

100

110

120

130

0

40

80

120

160

200

-4P

)

Liquid temp. (tc6)

Viscosity

Tem

pera

ture

(°C

)

Vis

cosity (10

0 1000 2000 3000 4000

Time (s)


4000


70

80

90

100

110

120

130

Vapour temp. (tc1)

Pressure (P2)

Liquid temp. (tc6)

Inhibitor injected

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pre

ssure

(bara

)

700 800 900 1000 1100

Time (s)



70

80

90

0

5

10

15

100

110

120

130

tc6 tc4 tc3

tc2

tc1

Vapour temps.

Liquid temps.

Inhibitor injector

pressure (P5)

Tem

pera

ture

(°C

)

Pre

ssure

(barg

)

850 900 950 1000

Time (s)


APPENDIX D LABORATORY MIXING TRIALS – FIGURES

t=2

.0 s

ec

t=3

.0 s

ec t

=6

.0 s

ec

t=

0.0

sec

t=

1.0

sec

t

=4

.0 s

ect=

15

.0 s

ec

inje

ctio

n

inje

ctio

n

Fig

ure

D1:

Com

parison B

etw

een the S

imula

tion a

nd E

xperim

ent R

esults (

Pitched

Bla

de T

urb

ine,1

00 r

pm

and B

ase Inje

ction)

inje

ctio

n

po

int

N=

10

0 r

pm

Top

in

ject

ion

aft

er 2

.0 s

ec

1

0.9

0.8

0.7

0.6

0.5

Vol

fra

ctio

n 0

.4

0.3

0.2

0.1

0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

To

p

Base

Tim

e(se

c)

Vo

lum

e o

ccu

pie

d b

y t

race

r u

sin

g a

cu

t-o

ff p

oin

t

of

10

% o

f fi

na

l m

ixed

co

nce

ntr

ati

on

Base

inje

ctio

n

po

int

N=

10

0 r

pm

inje

ctio

n a

fter

2.0

sec

Fig

ure

D2:

Com

parison o

f th

e T

racer

Dis

positio

n fro

m T

wo D

iffe

rent In

jection P

ositio

ns

t=

10

.0 s

ec

Inj:

t=

1.0

sec

t=

2.0

sec

t=3

.0 s

ec

t=

4.0

sec

t=

6.0

sec

t

=8

.0 s

ec

Fig

ure

D3:

Effect of th

e Inje

ction D

ura

tion o

n M

ixin

g T

ime (

Pitched

Bla

de T

urb

ine, 100 r

pm

)

Inj

1.8

s

1.0

s

Inj

3.8

s

t

= 1

0.0

sec

t= 0

.0

t= 1

.0 s

ec

t =

2.0

sec

t =

8.0

sec

Inje

ctio

n a

fter

Hal

f posi

tion

Zer

o s

ec

del

ay

Tw

o s

ec

del

ay

t

= 3

.0 s

ec

t

= 4

.0 s

ec

t =

5.0

sec

Impel

ler

stopped

wit

h d

elay

:

Inje

ctio

n p

osi

tio

n:

2 c

m f

rom

bo

tto

m

3 c

m f

rom

sh

aft

Imp

elle

r ty

pe

: R

ush

ton(1

/3)

Init

ial

imp

elle

r vel

oci

ty:

100

rp

m

t

= 2

.0 s

ec

t =

4.0

sec

t

=6.0

sec

t =

8.0

sec

t

= 1

0.0

sec

t =

12 s

ec

t

= 2

0.0

sec

t =

30.0

sec

Fiv

e se

c

del

ay

t =

5.0

sec

t =

10

.0 s

ec

t =

15

.0 s

ec

t =

20

.0 s

ec

t =

25

.0 s

ec

t =

30

.0 s

ec

t =

40

.0 s

ec

t

= 5

0.0

sec

Fig

ure

D4:

Tra

cer

Inje

ction b

y D

iffe

rent D

ela

y T

imes a

fter

Impelle

r S

tops (R

ushto

n , 1

00rp

m B

efo

re Im

pelle

r F

ailu

re)

Th

e m

ixin

g

beh

av

iou

r a

t th

e

sam

e (V

/Q)

for

dif

fere

nt

stir

rer

spee

d (p

itch

ed

bla

de

turb

ine)

Sp

eed

:

50 r

pm

t

=1.0

sec

t

=3.0

sec

t

=6.0

sec

t=

9.0

sec

t=

12.0

sec

t=

15 .0

t=0

.5 s

ec

t=

1.5

sec

t=

3.0

sec

t

=4

.5 s

ec t

=6

.0 s

ec

t=

7.5

sec

Sp

eed

:

150

rp

m

Sp

eed

:

100

rp

m

t=0.3

3 s

ec

t

=1.0

sec

t=

2.0

sec

t

=3.0

sec

t=

4.0

sec

t=

5.0

sec

Fig

ure

D5:

The T

racer

Spre

adin

g V

iew

s a

t V

arious F

raction a

nd M

ultip

le o

f (V

/Qp)

for

Diffe

rent Im

pelle

r S

peeds (

Pitched B

lade T

urb

ine)

Wit

h

Wit

ho

ut

ag

itati

on

Ag

ita

tio

n

10

0 r

pm

t

=0

.0 s

t=

0.2

s

t

=0

.4 s

t=

0.6

s

t=

0.8

s

t=1

.0 s

Fig

ure

D6: C

om

parison B

etw

een the T

racer

Spre

ad in

Agitate

d a

nd N

on-A

gitate

d C

ases a

t th

e S

am

e Inje

ction P

ressure

(R

ushto

n, T

op Inje

ctio

n, 10

0 r

pm

)

P=

14.7

psi

psi

psi

P=

7.1

P=

3.4

t

=0

.0 s

t=

0.2

s

t

=0

.4 s

t=

0.6

s

t=

0.8

s

t=1

.0 s

Fig

ure

D7: T

he T

racer

Spre

ad D

uring Inje

ction a

t V

arious P

ressure

s (N

o A

gitation, T

op Inje

ction)

Reaction Inhibition in theControl of Exothermic Runaway

Ref : JC 0400065

VISCOSITY ANNEX :

Development and Testing of an In-situ Viscosity Measuring System for the Adiabatic Dewar Calorimeter

Prepared by:

S.M. Rowe BSc, PhD, CChem and K.V. Middle BSc, CEng Chilworth Technology Ltd

TABLE OF CONTENTS

V1. INTRODUCTION ...........................................................................................................1

V1.1 IN-SITU VISCOSITY MEASUREMENT EXPERIMENTS ..........................................2

V2. DEVELOPMENT OF AN IN-SITU VISCOMETER ........................................................3

V2.1 SUPPLIER RESEARCH – SOFRASER FRANCE.....................................................5

V2.2 VISCOMETER SYSTEM DESIGN.............................................................................7

V2.3 COMMISSIONING THE SYSTEM .............................................................................8

V3. PERFORMANCE TESTING OF THE IN-SITU VISCOMETER ...................................13

V3.1 CALIBRATION TEST RESULTS .............................................................................13V3.1.1 Viscometer Calibration......................................................................................13V3.1.2 Modified Dewar Heat Capacity .........................................................................13

V3.2 EXPERIMENTAL TESTING ON THE POLYMERISATION OF STYRENE .............14V3.2.1 Standard Styrene Runaway Reaction...............................................................14V3.2.2 5% Polystyrene Runaway Reaction..................................................................14V3.2.3 10% Polystyrene Runaway Reaction................................................................15V3.2.4 20% Polystyrene Runaway Reaction................................................................15V3.2.5 Inhibitor Injection at 130°C (10% Polystyrene) .................................................16V3.2.6 Inhibitor Injection at 150°C (10% Polystyrene) .................................................16V3.2.7 Slow Inhibitor Injection at 150°C (10% Polystyrene) ........................................17V3.2.8 Inhibitor Injection at 150°C (10% Polystyrene) without Agitation......................18

V3.3 DISCUSSION OF TEST RESULTS .........................................................................27V3.3.1 Uninhibited Polystyrene / Styrene Runaway Reaction Data.............................27V3.3.2 Inhibited Polystyrene / Styrene Runaway Reaction Data .................................30V3.3.3 Effect of Varying Injection Conditions ...............................................................31

V4. CONCLUSIONS ..........................................................................................................33

Vii

V1. INTRODUCTION

The prediction of the performance of large scale injection systems based on small scale adiabatic

calorimetry studies requires accurate knowledge of a wide variety of chemical and physical

characteristics of the system under investigation. The conclusions from the initial phase of the

Reaction Inhibition (RI) project suggest that the more critical (and less easily defined) of these

parameters are :

x the effect of jet mixing (which predominates in small scale studies but will have lesser, but

potentially significant, impact under large scale mixing conditions), and

x the accurate definition of the viscosity of the reacting system throughout the course of the

injection and subsequent intermixing process.

At the present time, there are no known (commercially available) techniques for the co-incident

measurement of viscosity under adiabatic (runaway reaction) conditions. Viscosity can be

approximated through a variety of techniques including:

x Monitoring of relative torque changes in agitator systems

x Sampling of the calorimeter contents throughout a reaction and subsequent use of conventional

viscometer systems

x Conducting blowdown trials from a pressurised calorimeter to measure the flow of fluids through

calibrated pipework (only able to measure a single condition)

x Extrapolation from physical properties data (if the components of the reaction system including

intermediates and products are well defined and if suitable viscosity data exists).

However, all of these options are associated with appreciable uncertainty and can rarely be accurately

applied.

For the specification of a large scale reaction inhibition system, modelling of the injection and

subsequent intermixing processes requires precise knowledge of the viscosity of reaction system.

There are a range of conditions for which the RI system would have to function effectively (eg. the

presence or absence of mechanical agitation, the presence or absence of jet mixing effects, injection at

various stages through the runaway reaction, etc). To confirm the adequacy of RI for all of these

conditions, accurate viscosity data is vital. The requirement for this data is particularly pertinent in the

study of reactions that exhibit significant viscosity changes. Polymerisation reactions present a real

challenge in this respect and are also one of the major intended chemical processes to which RI may

be applied.

An extension to the original RI project was made possible by the sponsorship of ISPESL (the Italian

Safety Authority). The aims of the project extension were to:

x Research, develop, design and implement a method of viscosity measurement under the

demanding conditions of a runaway reaction

x Integrate the viscosity measuring system into the existing adiabatic Dewar calorimeter system

(with minimal impact on the functioning of the calorimeter)

x Demonstrate the viscosity measuring system on the vehicle reaction previously studied for the RI

project and examine the impact (on small-scale studies) of the viscosity of the reaction mass at the

point of inhibitor injection.

The adiabatic Dewar calorimeter system can be considered to be a perfectly mixed environment for

the study of inhibitor injection. In the context of the current project extension, this is advantageous (in

that the effects due to injection and mixing can be considered to be fully dissociated). However, for

large scale systems, the mixing conditions will be far from perfect and the effect of viscosity on the

inhibitor delivery rate (and subsequent mixing time) will be highly significant.

V1

V1.1 IN-SITU VISCOSITY MEASUREMENT EXPERIMENTS

Following successful implementation of a viscosity measuring system, a series of experimental trials

have been performed to evaluate the performance of the viscometer and to examine the effect of

viscosity on the mixing and distribution of inhibitor. Specifically, the following trials have been

performed:

Styrene polymerisation reaction initiated at 70°C and catalysed with 0.5 % w/w benzoyl peroxide. This

test was conducted to evaluate the performance of the modified Dewar calorimeter system compared

with the standard unit.

x Styrene / polystyrene (95% / 5% w/w) reaction initiated at 70°C and catalysed with 0.475 % w/w

benzoyl peroxide


benzoyl peroxide


benzoyl peroxide


benzoyl peroxide, with p-tert-butyl catechol (ptbc) inhibitor (0.01 molptbc.molstyrene -1) injected at

130°C



150°C



150°C without agitation



150°C with a prolonged injection period.

V2

V2. DEVELOPMENT OF AN IN-SITU VISCOMETER

Extensive research was conducted to source a viscometer sensor that :

x was small enough to fit inside a 1 litre adiabatic Dewar vessel

x has minimal impact on the thermal inertia of the adiabatic calorimeter (to retain phi factor (thermal

inertia) values consistent with the nature of the calorimeter)

x was sufficiently resilient to resist pressures of up to 30 barg and temperatures of up to 350°C

whilst retaining accuracy.

A thorough World Wide Web search was carried out for possible manufacturers of viscometers that

would meet the projects requirements. Drawings of the adiabatic Dewar installation were submitted to

various selected companies so that the problems associated with temperature, pressure and size

limitations could all be addressed directly with the relevant manufacturers.

Figure V2.1 shows the standard adiabatic Dewar calorimeter (ADCII) assembly and physical size of

the adiabatic shield internally. Figure V2.2 shows the plan view of the ADCII ported head. It can be

seen that sourcing a small enough viscometer for top entry into the standard vessel would be

challenging. It was therefore considered important, in the first instance, to investigate the availability

of the smallest viscometer possible and then address the temperature and pressure criteria issues

afterwards.

Overall, Chilworth Technology located at least 18 different viscometer manufacturers before selecting

the one described within this report. Many of their products were found unsuitable for the project due

to either the principle of operation (e.g. rotational viscometers), physical size, temperature rating,

pressure rating, viscosity range and other influencing factors relating to the intended test reactions.

Practicalities of cleaning the sensor between tests also played a major part in the assessment of

possible viscometers.

V3

Stirrer motor

Automatic vent

Back-up Rupture disc

Ported Head assembly

Stirrer coupling

shown)

300mm

513 m

m

Adiabatic shield

Anchor stirrer (heater not

Adiabatic Dewar vessel

Figure V2.1 Chilworth Technology standard Adiabatic Dewar Calorimeter installation.

Figure V2.2 Chilworth Technology standard ADC II head (shows amount of space available for a top entry viscometer).

V4

V2.1 SUPPLIER RESEARCH – SOFRASER FRANCE

Supplier research highlighted that Sofraser in France manufacture perhaps the smallest viscometer

transducer in the world of its type (the smallest that could be found in the extensive search of

manufacturers conducted by Chilworth). The Sofraser MIVI sensor is shown in Figure V2.3.

The MIVI sensor was capable of withstanding 300°C and 60 barg and therefore the sensor electronics

could potentially be installed within an oven.

The viscosity overall range of the sensor is dependent on the rod length and of the internal tuning

capacitor and the overall properties of each sensor; this sensor worked on the vibration principle. The

higher the damping of the vibration (and hence frequency), the higher the fluid viscosity. An important

aspect of the sensor design was that normal Dewar system agitation should not affect the measured

viscosity. This potential problem does not exist with the Sofraser MIVI sensor which has an imposed -1shear rate of 1000 – 2000 s compared with the agitator speed of up to 300 rpm.

It was decided that this viscometer had the most desirable overall capability although a modified

Dewar vessel would be necessary to house the sensor (due to the fact that the sensor length is 80 mm

and all of the sensor has to be immersed in the sample). Methods of installing this viscometer into a

Dewar system were investigated, with Figure V2.4 and Figure V2.5 showing some of the work

performed.

V5

120mm

70mm length

60mm dia

26mm thread dia

3mm dia

Figure V2.3 Sofraser MIVI sensor

Figure V2.4 Side entry of MIVI sensor to allow maximum probe coverage

V6

Figure V2.5 MIVI sensor – bottom entry to allow maximum probe coverage and with

adequate stirring action surrounding the probe.

Figure V2.4 illustrates the difficulty of side entry owing to the fact that the lower portion of the sensor

would be located in a poorly agitated zone.

V2.2 VISCOMETER SYSTEM DESIGN

Based on the detailed investigation of potential sensors, it was concluded that the Sofraser MIVI

sensor provided the best overall capability. It was, however, necessary to fabricate a custom

manufactured stainless steel Dewar vessel of wider neck diameter design. Figure V2.5 was the chosen

design (which also shows the gate stirrer design). Because the neck of the vessel would have to be

wider, to allow for gate stirrer removal between tests, a custom manufactured head and head clamp

were required. Figure V2.6 shows the custom design of the vessel.

V7

Figure V2.6 Custom manufactured vessel design with special gate stirrer. Connection at

bottom for accepting entry of viscometer sensor.

It was also noted that the new custom manufactured vessel and viscometer sensor would not fit

directly within the Chilworth standard adiabatic shield (oven) due to the internal chamber height

restriction. Therefore, it was decided that in order to align the vessel head automatic vent valve with

the existing pneumatic piston it would be required that a large hole be drilled in the bottom of the

oven, the vessel/viscometer lowered into the hole, the vent valve attached to the pneumatic arm and

the hole insulated to minimise heat losses around the viscometer to outside the oven. This was

therefore carried out and all the above parts ordered, including the viscometer sensor and processor

electronics control unit.

V2.3 COMMISSIONING THE SYSTEM

All system components were received satisfactorily together with the MIVI sensor. The sensor had

been pre-calibrated to standard Newtonian fluids in the range 0 - 4000 cP (although the sensor was

capable of measuring viscosities up to 10,000 cP). The calibration of the unit was bench checked prior

to installing into the custom manufactured vessel and proved to be very accurate. However, when the

viscometer was installed on to the new vessel and inserted within the oven it was noted that the zero

reading was unstable. Detailed investigation of the lack of stability indicated that the problem was

V8

related to the critical mass of the vessel. Previous industrial systems in which the MIVI sensor had

been installed had a minimum mass of 7 kg. This was vastly greater than the mass of the new vessel in

the current design.

To overcome this problem a robust clamping system was required. A special band clamp (with

eyelets) was manufactured. Eyelets were riveted in each corner of the oven interior base. A tubular

stainless section was also fabricated. This was used to seat the vessel and the viscometer passed

through the centre of this tube. Rigging screws were then used from each clamp eyelet to each oven

corner eyelet and tightened. This had the effect of tightening down the new vessel against the large

tubular mass at the oven base (centrally located) with the result of increasing the overall mass of the

new vessel. The viscometer was again tested and the zero-in-air reading rechecked. The reading was

found to be very stable and further testing to evaluate the performance of the system was conducted.

Figure V2.7 shows the MIVI sensor prior to installation within the new project vessel.

Figure V2.7 MIVI sensor

Figure V2.8 and Figure V2.9 show photos of the newly designed and manufactured vessel head and

stirrer assembly for the project.

V9

Figure V2.8 New wider head, with head stirrer assembly, custom manufactured gate stirrer

and heater.

Figure V2.9 New head design plan view

V10

Figure V2.10 Viscometer system

Figure V2.11 Viscometer system (prior to solving critical mass issues)

Figure V2.11 shows the set-up within the adiabatic shield prior to establishing that there were critical

mass issues associated with this set-up which would result in the “zero-in-air” instability initially

experienced.

V11

Figure V2.12 Viscometer system (shows solution to critical mass issues)

The MIVI sensor electronic processor (not shown) was located on a remote workbench and the

viscosity measurements were re-transmitted via a 4 - 20 mA link to the existing laboratory Dewar

acquisition software. The viscometer cables (not shown) are shrouded within a heat proof flexible

conduit and passed through the base of the oven.

V12

V3. PERFORMANCE TESTING OF THE IN-SITU VISCOMETER

Following the identification of an appropriate in-situ viscosity measuring system and the specification

and construction of a fully modified Dewar vessel and assembly to incorporate the measuring device, a

series of adiabatic Dewar tests have been undertaken to evaluate the performance of the modified

calorimeter.

Prior to commencement of the styrene tests, a series of calibration trials were performed to evaluate

the heat capacity of the modified Dewar system and calibrate the viscometer against a Newtonian

calibration fluid. Additional tests were conducted to confirm that the Dewar agitator speed (or internal

Dewar pressure) did not affect the measured viscosity.

V3.1 CALIBRATION TEST RESULTS

V3.1.1 Viscometer Calibration

The viscometer was calibrated using N1000 Viscosity Oil (for which a temperature / viscosity

relationship was provided by the supplier). Trials performed at different agitator stirring rates (100 –

300 rpm) showed clearly that the agitator speed did not affect the measured viscosity. However, for

the viscous oil tested, it was observed that appreciable heat input occurred at higher rates of stirring -1(ie. 0.17 K.min at 30°C for a stirrer speed of 300 rpm). This is not an unexpected finding for the high

viscosity fluid used (2200 cP at 30°C) but does illustrate the potentially significant effect of stirrer heat

input in the adiabatic study of high viscosity systems.

The reproducibility of the viscosity measurements (following disassembly / re-assembly of the Dewar

system) was initially found to be very poor (completely different results were noted each time the

equipment was re-assembled). As previously explained, this was attributable to the critical mass of the

Dewar assembly. The system was modified to incorporate substantial bracing (a strong retaining

clamp system (bolted to the oven floor) with a modified circular stand on which the system was

supported). This modification provided a system which was reliable and reproducible between tests

(without the need for re-linearisation of the viscometer prior to every test). Subsequent tests were

performed on the modified system. The high shear rate of the viscometer (1000 – 2000 s-1 was

sufficiently in excess of the normal range of stirrer speeds (up to 300 rpm) that interference of the

agitator with the measured viscosity was not observed when trials were performed at various stirrer

speeds.

V3.1.2 Modified Dewar Heat Capacity

Step-wise heating experiments were performed using 800 g of water in the range 35 to 90°C (and with

a heating rate of 80 W). The mean (static) heat capacity determined for the Dewar vessel was found to

be 620 J.K-1. This equates to a phi factor (for the styrene / polystyrene reaction system) of 1.379. The -1heat capacity determined is appreciably higher than that of the standard Dewar system (250 J.K

which equates to a phi factor of 1.15).

The cause of this appreciable increase is believed to be two-fold. Firstly, the modified Dewar system is

appreciably heavier than the standard system (predominantly due to the inclusion of the bulky

viscometer). Secondly, and most significantly, the inclusion of the viscometer into the base of the

Dewar system means that there is a large un-protected surface area on the base of the vessel through

which heat can transfer from the inside to the outside of the vessel.

With the viscometer intimately attached to the base of the vessel (and protruding through the base of

the oven) it is observed that the isothermal temperature stability of the Dewar system is poor in

relation to the standard system. During the isothermal hold periods following heat capacity

V13

-1calibrations, a downward drift in temperature of around 0.07 K.min (4.1 K.hr-1) was observed at

90°C, despite the heat input of the agitator previously discussed. This value would be expected to

increase at higher temperatures as the differential temperature between the sample and the outside of

the oven increases. This significantly impacts on the adiabaticity of the system and requires

appreciable attention as further work.

Despite the short-comings in the adiabaticity of the Dewar system, further trials to examine the

efficacy of the system under runaway conditions were conducted. The reaction vehicle selected for

study was the same system previously examined in the RI project (i.e. the peroxide catalysed

polymerisation of styrene). In order to evaluate the efficiency of the in-situ viscosity system, the

original styrene formulation was modified by the addition of polystyrene (an amorphous polymer with

an average molecular mass of ca. 230,000 (average molecular number ca. 140,000); glass transition

temperature of 94°C). An experimental programme of work was performed to evaluate the efficacy of

the viscometer in the presence of different concentrations of polystyrene. In addition, trials were

performed to evaluate the effect of inhibitor injection to the more viscous formulation.

V3.2 EXPERIMENTAL TESTING ON THE POLYMERISATION OF STYRENE

V3.2.1 Standard Styrene Runaway Reaction

Date of Test : 10th September 2002

Filename : D10402P/Q.ASC

Operator : P.J. Carter

Data currently exists on the styrene (BPO catalysed) runaway reaction initiated from 70°C using the

standard Dewar system. The first test in the modified system was performed to compare the data from

the standard test with that from the modified test system (whilst also recording the viscosity of the

system).

Styrene (696.5 g) and benzoyl peroxide (3.5 g (of dry material)) were charged to the modified Dewar

system. The internal Dewar heater was employed to heat the reaction mass up to 70°C at which point

the reaction was maintained under adiabatic conditions. Figure V3.1 illustrates the temperature /

pressure / viscosity / time data obtained. After reaching 70°C, the reaction accelerated (slowly at first) -1reaching maximum conditions of 306.5°C and 7.0 barg at maximum rates of 53.5 K.min and 3.5

-1bar.min . The time to maximum rate was noted to be 67.9 minutes.

The viscosity began to increase detectably from ca. 150°C with the most rapid rate of viscosity

increase observed at around 260°C. The peak viscosity noted was 1908 cP at 302°C (this was observed

4 minutes after the main exothermic reaction was noted to be complete).

V3.2.2 5% Polystyrene Runaway Reaction


Filename : D10402R/S.ASC


The standard pure styrene formulation was modified to incorporate 5% polystyrene (with a

corresponding reduction in the BPO concentration to retain the same ratio of styrene to BPO).

Styrene (665 g), polystyrene (35 g) and benzoyl peroxide (3.325 g (of dry material)) were charged to

the modified Dewar system. The internal Dewar heater was employed to heat the reaction mass up to

70°C at which point the reaction was maintained under adiabatic conditions.

V14

Figure V3.2 illustrates the temperature / pressure / viscosity / time data obtained. After reaching 70°C,

the reaction accelerated (slowly at first) reaching maximum conditions of 297.8°C and 6.0 barg at -1 -1maximum rates of 44.8 K.min and 2.81 bar.min . The time to maximum rate was noted to be 72.8

minutes.


increase observed at around 270°C. At the end of the exothermic heat release, the viscosity was 1661

cP although over the following 35 minutes, this increased appreciably to 3730 cP.



Filename : D10402T/U.ASC








the reaction accelerated (slowly at first) reaching maximum conditions of 288.9°C and 5.0 barg at -1 -1maximum rates of 40.8 K.min and 2.28 bar.min . The time to maximum rate was noted to be 92.1

minutes.


increase observed at around 270°C. At the end of the exothermic heat release, the viscosity was ca.

3000 cP although over the following 15 minutes, this increased appreciably to 5432 cP.

In a preliminary test on this formulation, a significant increase in viscosity was noted to occur from

210°C which reached 8500 cP within 1 minute. The viscosity remained constant through the

subsequent exothermic reaction decreasing sharply on reaching the peak temperature. The remainder

of the viscosity profile after this discontinuity was essentially the same as that noted for the repeat test.

The cause of this discontinuity is discussed in more detail in the discussions section of this report

(section V3.3). In summary, it is believed to have been caused by localised cooling of the reaction

mixture in the base of the calorimeter (where the heat losses are highest) causing partial solidification

of the mass around the viscometer.


Date of Test :

Filename :

Operator :

13th June 2002 D10402K/L.ASC H.E. Gair






V15


the reaction accelerated (slowly at first) reaching maximum conditions of 268°C and 4.5 barg at -1 -1maximum rates of 33.7 K.min and 1.78 bar.min . The time to maximum rate was noted to be 77.1

minutes.


increase observed at around 240°C. The viscometer reached its limiting value at 250°C (hence

viscosity data above this point could not be determined).

This test was performed prior to the attachment of the enhanced securing devices (and re-calibration of

the sensor from 0 to 8000 cP (rather than the initial setting of 0 to 4000 cP). These additions were

noted to cause significant additional heat loss from the base of the calorimeter. The time to maximum

rate in the current test is therefore less than that for the 10% polystyrene sample (test V3.2.3).

V3.2.5 Inhibitor Injection at 130°C (10% Polystyrene)


Filename : D10402V/W.ASC



corresponding reduction in the BPO concentration to retain the same ratio of styrene to BPO). In

addition, the effect of a 1% mol/mol injection of p-tert-butyl catechol inhibitor was assessed.

Styrene (605 g), polystyrene (70 g) and benzoyl peroxide (2.8 g (of dry material)) were charged to the

modified Dewar system. The internal Dewar heater was employed to heat the reaction mass up to

70°C at which point the reaction was maintained under adiabatic conditions. On reaching 130°C, ptbc

(10.0 g) in styrene (25 g) was injected into the calorimeter from an air-pressurised bomb (to ensure

rapid injection in < 1 s). The bomb volume of 75 cm3 and a 6 barg air pressure was used to force the

injection.


the reaction accelerated (slowly at first) reaching 130°C after 80 minutes. At this point, the inhibitor

injection was performed causing a decrease in the sample temperature (owing to the addition of the

cold inhibitor solution). From this point, the sample temperature continued to decrease (ultimately

reaching 73°C some 17 hours after the injection).

It is believed that the extremely slow rate of residual heat release after inhibitor injection was of lesser

magnitude than the heat loss from the system caused by the attachment of the viscometer. The

viscosity of the system was 73 cP immediately after the injection increasing to 180 cP during the

cooling period to 73°C. A further test was performed with an injection temperature of 150°C.

V3.2.6 Inhibitor Injection at 150°C (10% Polystyrene)


Filename : D10402X/Y.ASC




addition, the effect of a 1% mol/mol injection of p-tert-butyl catechol inhibitor was assessed.

V16




(10.0 g) in styrene (25 g) was injected into the calorimeter from an air-pressurised bomb (to ensure 3rapid injection in < 1 s). The bomb volume of 75 cm and a 6 barg air pressure was used to force the

injection.


the reaction accelerated (slowly at first) reaching 150°C after 86 minutes. At this point, the inhibitor

injection was performed causing a decrease in the sample temperature (owing to the addition of the

cold inhibitor solution). From this point, the sample temperature continued to decrease to 109°C over

20 hours before starting to increase once more. Over the following 75 hours, the temperature increased

to 175°C. The test was terminated at this point as the viscosity had reached a high level (7400 cP) and

the rate of temperature rise was very low.

It is believed that the extremely slow rate of residual heat release after inhibitor injection was of lesser

magnitude than the heat loss from the system caused by the attachment of the viscometer. However,

over the 20 hours taken to reach 109°C, the activity of the inhibitor had reduced to such an extent that

the power output from thermal polymerisation was increasing significantly (this aspect is discussed in

Section 4.2 of the main report). The rate of heat release at this point began to exceed the rate of heat

loss from the calorimeter and slow self-heating occurred. The viscosity was successfully measured

throughout the test.

V3.2.7 Slow Inhibitor Injection at 150°C (10% Polystyrene)

Date of Test : 6th November 2002

Filename : D10402cc.ASC




addition, the effect of a 1% mol/mol injection of p-tert-butyl catechol (a known inhibitor) was

assessed. As opposed to test 3.2.6, the inhibitor solution was pumped into the calorimeter over 2

minutes rather than being injected instantaneously using gas pressure.




(10.0 g) in styrene (25 g) was injected into the calorimeter via a high pressure pump over a total

duration of 2 minutes.

Figure V3.7 illustrates the temperature / pressure / viscosity / time data obtained.

After reaching 70°C, the reaction accelerated (slowly at first) reaching 150°C after 100 minutes. At

this point, the inhibitor injection was performed over a 2 minute duration. The sample temperature

showed evidence of slowing at 168°C before finally stabilising at 180°C. The sample temperature

continued to rise (albeit slowly) ultimately reaching a peak temperature of 203°C (3.7 barg) some 103

minutes after commencement of the addition. The sample temperature then essentially stabilised

remaining between 194 and 214°C over the following 20 hours. The test was terminated at this point

as the viscosity had reached a high level (7550 cP) and the rate of temperature rise was very low.

The viscosity had increased steadily following completion of the addition (at which point the

measured viscosity was 418 cP). The slow addition clearly causes a delay in stabilising the increasing

V17

3

temperature. Once stabilisation was achieved (at 180°C) the rate of continuing reaction was

sufficiently high to overcome the modified calorimeter heat losses.

V3.2.8 Inhibitor Injection at 150°C (10% Polystyrene) without Agitation

Date of Test : 23rd October 2002

Filename : D10402aa.ASC




addition, the effect of a 1% mol/mol injection of p-tert-butyl catechol (a known inhibitor) was

assessed. As opposed to test V3.2.6, the inhibitor solution was added via an air-pressurised bomb but

with the calorimeter agitator isolated at 130°C (such that nominally static conditions were present in

the vessel at the point of injection).



70°C at which point the reaction was maintained under adiabatic conditions. On reaching 130°C, the

agitator was switched off and at 150°C, ptbc (10.0 g) in styrene (25 g) was injected into the

calorimeter from an air-pressurised bomb (to ensure rapid injection in < 1 s). The bomb volume of 75

cm and a 6 barg air pressure was used to force the injection. The inhibitor was injected into the

headspace of the vessel (ie. not sub-surface).

Figure V3.8 illustrates the temperature / pressure / viscosity / time data obtained.

After reaching 70°C, the reaction accelerated (slowly at first) reaching 150°C after 86 minutes. The

agitator was isolated at 130°C. At 150°C, the inhibitor injection was performed but, due to the absence

of agitation, no appreciable decrease in the rate of reaction was observed until the temperature had

reached 163°C. After a short settling period, the temperature began to increase once more reaching a

peak of 249.3°C (2.8 barg) some 3.2 hours after the injection had been conducted. Having reached the

peak value, the temperature began to subside (due to heat losses) with the test finally terminated 16

hours after completion of the injection. The viscosity had initially decreased for ca. 60 minutes after

the injection (reaching a minimum value of 1381 cP) before increasing steadily throughout the

duration of the test reaching the limit of the sensor (10000 cP) 7 hours after the injection.

The absence of agitation clearly has an adverse effect on the initial intermixing process. In this case,

the temperature is seen to rise an additional 13 K above the injection point before the temperature

increase is halted. This in turn leads to a more rapid recovery after injection. The magnitude of the

effect of agitation is likely to be much more appreciable in larger scale vessels where the initial jet

mixing effect of the injected stream will be less significant.

V18

Figure V3.1 Experimental Data from Pure Styrene / BPO Runaway

0

40

80

120

160

200

240

280

320

0

1

2

3

4

5

6

7

8

Tem

pera

ture

(°C

)

Pre

ssure

(barg

)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (hours)

Sample Temperature Pressure

100

120

140

160

180

200

220

240

260

280

300

320

Tem

pera

ture

(°C

)

0

500

1000

1500

2000

2500

Vis

co

sity (

cP

)

95 97 99 101 103 105 107 109 111 113

Time (minutes)

Sample Temperature Viscosity

V19

Figure V3.2 Experimental Data from Styrene / Polystyrene (5%) / BPO Runaway

0

50

Tem

pera

ture

(°C

)

0

1

2

3

4

5

6

7

Pre

ssu

re (

ba

rg)

100

150

200

250

300

0 20 40 60 80 100 120 140 160 180

Time (minutes)


0

50

100

150

200

250

300

Tem

pera

ture

(°C

)

0

Vis

co

sity (

cP

)

500

1000

1500

2000

2500

3000

3500

4000

110 120 130 140 150 160 170

Time (minutes)


V20


0

50

100

150

200

250

300

Tem

pera

ture

(°C

)

0

1

2

3

4

5

6

Pre

ssure

(barg

)

0 20 40 60 80 100 120 140 160

Time (minutes)


0

50

0

Vis

cosity (

cP

)

100

150

200

250

300

Tem

pera

ture

(°C

)

1000

2000

3000

4000

5000

6000

7000

8000

125 130 135 140 145 150

Time (minutes)

Sample Temperature Viscocity

V21


0

50

Tem

pera

ture

(°C

)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Pre

ssure

(barg

)

100

150

200

250

300

3 3.5 4 4.5 5 5.5 6

Time (hours)


0

50

0

Vis

co

sity (

cP

)

100

150

200

250

300

Tem

pera

ture

(°C

)

1000

2000

3000

4000

5000

6000

4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6

Time (hours)


V22

Figure V3.5 Experimental Data from Styrene / Polystyrene (10%) / BPO Runaway with

Inhibitor Injection at 130°C

0

20

40

60

80

100

120

140

Tem

pera

ture

(°C

)

0

1

Pre

ssure

(barg

)

0.2

0.4

0.6

0.8

1.2

1.4

1.6

0 2 4 6 8 10 12 14 16 18

Time (hours)


20

40

60

80

0

20

40

60

80

Vis

cosity (

cP

)

100

120

140

Tem

pera

ture

(°C

)

100

120

140

160

180

200

0 2 4 6 8 10 12 14 16 18

Time (hours)


V23


Inhibitor Injection at 150°C

0

20

40

60

80

Tem

pera

ture

(°C

)

0

1

2

3

Pre

ssure

(barg

)

100

120

140

160

180

0.5

1.5

2.5

3.5

0 2 4 6 8 10 12 14 16 18 20

Time (hours)


0

20

40

60

80

100

120

140

160

180

Tem

pera

ture

(°C

)

0

Vis

cosity (

cP

)

500

1000

1500

2000

2500

3000

0 2 4 6 8 10 12 14 16 18 20

Time (hours)


V24


Slow Inhibitor Injection (over 2 minutes) at 150°C

0

20

40

60

80

Tem

pera

ture

(°C

)

0

1

2

3

4

Pre

ssure

(barg

)

100

120

140

160

180

200

220

0.5

1.5

2.5

3.5

0 2 4 6 8 10 12 14 16 18 20 22 24

Time (hours)


0

20

40

60

80

100

120

140

160

180

200

220

Tem

pera

ture

(°C

)

0

Vis

cosity (

cP

)

1000

2000

3000

4000

5000

6000

7000

8000

9000

0 2 4 6 8 10 12 14 16 18 20 22 24

Time (hours)


V25


Inhibitor Injection at 150°C (Agitator Switched off at 100°C)

0

50

Tem

pera

ture

(°C

)

0

1

2

3

Pre

ssu

re (

ba

rg)

100

150

200

250

300

-0.5

0.5

1.5

2.5

0 2 4 6 8

Time (hours)


0

20

40

60

80

100

120

140

160

180

200

220

240

260

0

2000

4000

6000

8000

10000

12000

Vis

co

sity (

cP

)

Tem

pera

ture

(°C

)

0 1 2 3 4 5 6 7 8 9

Time (hours)


V26

V3.3 DISCUSSION OF TEST RESULTS

V3.3.1 Uninhibited Polystyrene / Styrene Runaway Reaction Data

The in-situ viscometer system has proved highly effective in providing detailed viscosity analysis

during the progression of a runaway reaction. Unfortunately, the integration of the viscometer into the

Dewar calorimeter assembly substantially impacts on the thermal characteristics of the system. Data is

compared in Table V3.1 and

Figure V3.9 for the BPO catalysed polymerisation of styrene when conducted in the standard ADC II

and the viscometer-modified calorimeter.

Table V3.1 Comparison of Data for Standard and Modified Dewar Systems

Standard Dewar Viscometer Dewar

Tmax (°C) 339.6 306.6 Pmax (barg) 12.6 7.0 dT/dtmax (K.s-1) 1.16 1.09 dP/dtmax (bar.s -1) 0.094 0.059 Time to peak temperature from 70°C (minutes) 49 69 Phi Factor (calculated) 1.15 1.379

50

Tem

pera

ture

(°C

)

100

150

200

250

300

350

0 10 20 30 40 50 60 70

Time (hours)

Modified Dewar Data Standard Dewar Data

Figure V3.9 Comparison between Modified and Standard Dewar Data for Styrene

It is clear that all measured thermal parameters are adversely affected by the integration of the

viscometer. However, the magnitude of the effect is not sufficiently large to render the data -1meaningless. The theoretical phi factor (assuming a Dewar vessel heat capacity of 620 J.K ) for the

V27

modified Dewar vessel is calculated to be 1.379 whereas the “apparent phi factor” is observed to be

1.31 (this is the ratio between the known adiabatic temperature rise for the reaction and the measured

temperature rise for the modified Dewar). This anomaly is likely to be attributable to the variation in

the vessel heat capacity at different self-heat rates.

For the purposes of providing thermal data for detailed safety system design, the standard Dewar

configuration should be employed for the runaway characteristics. However, viscometry data obtained

from the modified system at the emergency relief system design temperature is also considered valid

for design purposes, since the viscosity data is likely to be marginally conservative owing to the higher

phi factor – at a given temperature in a higher phi factor apparatus, a higher conversion would have

occurred compared to the standard Dewar, leading to a higher, conservative, viscosity value.

The influence of initial polystyrene content on the viscosity of the reaction mass is clearly

demonstrated by tests V3.2.1 to V3.2.4. Data relating % conversion, temperature and viscosity are

provided in Table V3.2 and Figure V3.10 for these trials.

Table V3.2 Viscosity of Reaction Mixture

Temperature 0% PS

%PS Viscosity

5% PS %PS Viscosity

10% PS %PS Viscosity

20% PS %PS Viscosity

70°C 0 0 5 0 10 69 20 875 100°C 12.7 8 17.5 9 22.3 80 32.1 418 125°C 23.3 15 27.9 24 32.6 100 42.2 509 150°C 33.8 27 38.4 39 42.9 241 52.3 584 175°C 44.4 57 48.8 74 53.2 477 62.4 954 200°C 55.0 107 59.2 127 63.4 767 72.5 1173 225°C 65.5 176 69.6 294 73.7 1157 82.6 2457 250°C 76.1 274 80.1 512 84.0 1638 92.7 5250 275°C 86.7 920 90.5 828 94.3 2338 N/A N/A 300°C 97.3 1263 N/A N/A N/A N/A N/A N/A

Notes:

x %PS is the percentage of polystyrene in the reaction mixture during the polymerisation process.

This is calculated from the percentage of thermal conversion plus the initial polystyrene content.

x Viscosity units are cP

wAt initial polystyrene concentrations of 0 and 5% /w, there is little discernible difference in the

measured viscosities of the mixtures throughout the runaway. However, increasing concentrations to

10 and 20% w/w makes an appreciable difference to the measured values. At 20% w/w, it is seen that the

initial mixture is fairly viscous. The viscosity of this solution decreases in the early stage of the

runaway (indicating that the temperature effect dominates over the conversion effect).

The absolute viscosity values should be regarded as indicative although the calibration testing

conducted on the sensor suggests a tolerance within 10%.

V28

0

Vis

cosity (

cps)

1000

2000

3000

4000

5000

6000

50 100 150 200 250 300

Temperature (°C)

0% PS 5% PS 10% PS 20% PS

Figure V3.10 Viscosity versus Temperature Profile

Analysis of the rate of temperature rise profiles for the various initial polystyrene concentrations

examined reveals that the kinetic mechanism of the reaction remains unaltered by the presence of the

initial polystyrene (Figure V3.11). Three rate peaks are observed in all tests corresponding with two

stages of BPO initiation followed by pure thermal polymerisation. However, owing to the increased

loading of polystyrene, the magnitude of each peak is seen to diminish; that is the kinetics are slowed

due to concentration effects.

0

10

20

30

40

50

60

70

dT

/dt

(K/m

in)

70 120 170 220 270

Time (mins)

dT/dt (0% PS) dT/dt (5% PS) dT/dt (10% PS) dT/dt (20% PS)

Figure V3.11 Comparison of Rate of Temperature Rise for PS Formulations

V29

V3.3.2 Inhibited Polystyrene / Styrene Runaway Reaction Data

Injection of inhibitor into the runaway reaction of 10% polystyrene / styrene mixture at 130 and 150°C

provides effective inhibition of the runaway reaction. A comparison of temperature and viscosity data

for the inhibited reactions is provided in Figure V3.12.

The initial stages of the data (i.e. prior to inhibitor injection) demonstrate the reproducibility of the

thermal data from the modified Dewar system. Following execution of the (rapid) injection, immediate

cessation of reaction is observed. The increased viscosity of the reaction mass at the point of injection

(compared with the pure styrene runaway) does not adversely affect the subsequent inhibition of the

reaction. However, since the small scale Dewar system can be considered to be near-perfectly mixed,

this is not particularly surprising. The duration of intermixing at larger scales would be expected to

increase considerably with increasing viscosity.

The relatively poor thermal stability characteristics of the modified Dewar assembly are such that on

completion of the injection, the low rate of continued propagation is appreciably less than the heat loss

from the reactor. For example, with injection at 130°C, the rate of temperature rise associated with

continued reaction has previously been observed to be 10.8 K.hr-1. In the case of the modified Dewar

system, a temperature rise rate of – 49.5 K.hr-1 is observed, despite the heat input of the agitator into

the advancing viscosity of the polystyrene. The viscosity in the reactor after settling, is observed to

drift upwards very slowly whilst the reactor contents cools. Over the 17 hours post-injection, the

temperature decreases from 130°C to 73°C whilst the viscosity increases from 74 cP to 180 cP.

0

50

100

150

200

250

300

Tem

pera

ture

(°C

)

0

1000

2000

3000

4000

5000

6000

7000

8000

Vis

cosity (

cP

)

-80 -70 -60 -50 -40 -30 -20 -10 0 10 20

Time (minutes)

No Inhibitor (Temp) Inhibited (Temp) Inhibited (130°C, Temp) No Inhibitor (Visc) Inhibited (Visc) Inhibited (130°C, Visc)

Figure V3.12 Comparison of Data for Inhibitor Injection Studies

V30

Under large scale heat loss conditions (i.e. near adiabatic), it would be anticipated that the temperature

would remain essentially constant after injection whilst the viscosity would continue to rise slowly due

to continuing (albeit slow) reaction.

For injection at 150°C, the initial rate of temperature rise associated with continued reaction has

previously been observed to be 32.4 K.hr-1. In the case of the modified Dewar system, a temperature -1rise rate of – 4.5 K.hr is observed, despite the agitation effect. The viscosity in the reactor after

settling, is observed to drift upwards very slowly whilst the reactor contents initially cools (down to

109°C) and then increases. Over the 93 hours post-injection, the temperature reaches a peak of

175.4°C whilst the viscosity increases from 1700 cP to 7400 cP.

From initial inspection of the inhibitor injection studies, it is not possible to ascertain whether the

viscosity after injection is increasing due to the temperature decreasing or to a combination of falling

temperature and continuing conversion. Although there is an appreciable amount of data collected on

the uninhibited polymerisation, it has not proven possible to identify whether the viscosity change is

solely attributable to the temperature change due to the inherent variability in the data. Further studies

would be required to enable this type of data analysis.

V3.3.3 Effect of Varying Injection Conditions

The effect of varying injection conditions is seen to make an appreciable difference to the overall

efficacy of the inhibition process (

Figure V3.13). This is likely to be substantially affected by the increased viscosity of the batch at the

moment of injection, which will significantly reduce the rate of inhibitor diffusion.

Injecting into the (headspace of the) Dewar vessel with the agitator inactive gives rise to an overshoot

in the baselining temperature (i.e. the equilibrium temperature at which the reaction is brought under

control). The reaction is seen to be inhibited but the baselining temperature is sufficiently high that the

rate of continuing reaction is able to overcome the modified calorimeter heat loss (and hence self-

heating continues).

Provision of a relatively long (2 minutes) inhibitor delivery duration by pumping through a narrow

bore feed line into the headspace with continued agitation, similarly causes an overshoot in the

baselining temperature (by a higher margin). In this trial, jet mixing effects are likely to be absent and

hence the mechanical agitation is the sole means of inhibitor distribution. By contrast, with the

pressurised injection system used for all other trials, a significant contribution to the overall mixing

efficiency would be likely to result from the turbulence caused by the high rate injection below the

liquid level and the subsequent jet of pressurising gas. It is likely that during plant scale injection

studies, viscosity will have a pivotal role in the effectiveness of these jet mixing effects. At high

viscosity, the level of turbulence invoked by the (sub-surface) jet of inhibitor and gas is likely to

reduce considerably (hence diminishing the overall efficiency of the intermixing process).

V31

Te

mp

era

ture

(°C

)

190

185

180

175

170

165

160

155

150

145

140

-20 0 20 40 60 80 100 120

Time (seconds)

No Injection Injection at 150°C Slow Injection No Agitation

Figure V3.13 Comparison of Varying Inhibitor Injection Conditions

V32

V4. CONCLUSIONS

Integration of an in-situ viscometer into an adiabatic calorimeter unit has been successfully achieved

and demonstrated. The adiabatic Dewar calorimeter system is likely to be one of the only

commercially available adiabatic calorimeters in which such as system would be feasible, owing to the

relatively large test cell volume. Although the smallest possible viscometer sensor was sought, the size

of the smallest unit identified remained fairly substantial for laboratory use.

An appreciable number of technical obstacles were successfully overcome yielding a reliable and

reproducible calorimeter system with in-situ viscosity capability. The stability of the Dewar assembly

was found to be critical in overcoming problems associated with vessel vibration interfering with the

oscillation of the viscometer sensor.

As would be anticipated, deterioration in performance of the calorimetric function of the system was

noted due to the appreciable mass of the viscometer system. This performance deterioration was

exacerbated by enhanced heat losses from the calorimeter (caused by heat transfer through the main

body of the viscometer to the ambient temperature environment). The as-built modified Dewar suffers

significantly from heat losses from the base of the calorimeter, through the viscometer. The magnitude

and consequences of this effect were only identified once elevated temperature, low self-heat rate

reactions were tracked (i.e. during the inhibitor injection trials). This precluded the possibility of re-

design of the system during the study.

In order to enhance the performance of the system, future work will attempt to mount the entire

viscometer assembly within the oven (this will require a larger (taller) oven than standard). It may also

be necessary to modify the coupling of the viscometer to the Dewar vessel to reduce the rate of heat

loss to the body of the viscometer. This may decrease the phi factor (i.e. improve the thermal

response).

The viscometer unit works well and reliably. It is relatively easy to clean. The modifications to the

Dewar ensure that all normal operations remain possible (e.g. pumped additions, pressurised

injections, agitation options, etc). Further work should be undertaken to re-design the coupling system

to focus on improving the thermal characteristics of the vessel. On completion of this modification, a

further programme of experimental work should be performed to further evaluate the injection and

intermixing process in viscous systems. The current work has considered only a limited range of

injection conditions.

The use of the system in the styrene examples is clearly demonstrated. However, with the current

system, the kinetics and thermodynamics have been compromised to an unacceptable extent.

The significance of adding on-line viscometry to the normal functions of an adiabatic calorimeter

should not be underestimated. As well as providing crucial information of the design of a RI system,

viscosity data is vital in the specification of other basis of reaction safety. For example, in emergency

relief system design, standard vent sizing equations are only valid for vented fluids with a viscosity of

below 100 cP. Alternative calculation techniques exist for sizing relief systems for fluids that exceed

this viscosity. Clearly, the implications of transporting a viscous fluid through a relief vent system can

be appreciable. There is presently no known practical technique for confirmation of (a reacting) relief

stream viscosity under experimental conditions. The same critical requirement for viscosity data also

exists for quenching and drown-out techniques (both of which require rapid and reliable intermixing of

the relieved stream with the diluent).

For specification of large scale reaction inhibition systems, the viscosity of the fluid during runaway

will be a significant parameter, particularly for polymerising systems. Using a technique such as

Computational Fluid Dynamics (CFD), modelling of the inhibitor injection flow patterns and the

intermixing processes is possible to confirm adequate delivery and mixing times within the constraints

of the runaway reaction kinetics. CFD is required for its ability to model fluid dynamics, whereas the

V33

Network-of-Zones simulation is only able to simulate the concentration and reaction profiles once the

flow patterns are known. Therefore, for this application, on-line viscometry is crucially important in

the design of successful inhibition systems.

From initial pilot scale trials it is observed that jet mixing of the inhibitor into the (mobile) reaction

media provides rapid and effective intermixing and distribution of the inhibitor. With increasing

viscosity at the point of injection, the effect of this jet mixing will diminish appreciably. The impact of

this effect can only feasibly be considered through the conduct of CFD modelling of the large scale

injection process.

V34

Printed and published by the Health and Safety ExecutiveC30 1/98

Printed and published by the Health and Safety Executive C1.10 10/03

ISBN 0-7176-2730-6

RR 145

78071 7 627301£30.00 9

Date post:	07-Nov-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

RR 145 - Reaction Inhibition in the Control of Exothermic Runaway · Title RR 145 - Reaction...

Documents